The Faculty of Information Technology congratulates Dr. Le Ngoc Thanh on the successful defense of his Ph.D. thesis titled "Link Prediction on Knowledge Graphs using Transition-based and Convolutional Network-based Embedding". The research was supervised by Prof. Dr. Le Hoai Bac, Head of the Computer Science Department.
Dr. Le Ngoc Thanh's thesis proposed four solutions to improve the accuracy of link prediction on knowledge graphs, which are:
- Applying HyperNetwork in neural network-based models to create better filters for the process of mining interactions between entities and relations, while also initiating weights to help the network search faster and increase network depth;
- Implementing entity projection on hyperplanes for relation determination, combined with complex rotation;
- Exploiting Quaternion space and group theory in double and isotropic rotations to identify tail entities;
- Integrating convolution with rotation to enhance information in the transformation process. Each proposal has been tested on standard datasets and evaluated from multiple aspects to demonstrate the model's predictive effectiveness.
The Faculty of Information Technology wishes him good health and hopes he will continue to contribute to the teaching and research activities of the Faculty.